1 class GenscanReport:
   2     def __init__(self):
   3         self.genes = list()
   4     def addGene(self, gene):
   5         self.genes.append(gene)
   6 
   7 class Gene:
   8     def __init__(self, no):
   9         self.no = no
  10         self.exons = list()
  11         self.cds = ''
  12         self.peptide = ''
  13     def addExon(self, exon):
  14         self.exons.append(exon)
  15 
  16     def setCds(self, fasta):
  17         self.cds = fasta
  18     def setPeptide(self, fasta):
  19         self.peptide = fasta
  20 
  21 class Exon:
  22     def __init__(self, no, type, strand, begin, end):
  23         self.no = no
  24         self.type = type
  25         self.strand = strand
  26         self.begin = int(begin)
  27         self.end = int(end)
  28 
  29 class FastaIterator:
  30     def __init__(self, ifile):
  31         self.ifile = ifile
  32         self.g = self.getGenerator()
  33     def getGenerator(self):
  34         while True:
  35             line = self.ifile.next().strip()
  36             if line:
  37                 lines = [line]
  38                 break
  39         for line in self.ifile:
  40             if line.startswith('Explanation'):
  41                 yield ''.join(lines)
  42                 break
  43             if line.startswith('>'):
  44                 yield ''.join(lines)
  45                 lines = [line]
  46             else:
  47                 lines.append(line)
  48 
  49     def __iter__(self):
  50         return self.g
  51 
  52     def next(self):
  53         return self.g.next()
  54 
  55 def parseGenscan(handler):
  56     gr = GenscanReport()
  57     while True:
  58         line = handler.next()
  59         if line.startswith('-') or line is None: break
  60     i = 0
  61     gene = None
  62     for line in handler:
  63         line = line.strip()
  64         if line.startswith('P') or line.startswith('S'): break
  65         if not line:
  66             if gene: gr.addGene(gene)
  67             i+=1
  68             gene = Gene(i)
  69             continue
  70         words=line.split()
  71         gene.addExon(Exon(words[0], words[1], words[2], words[3], words[4]))
  72 
  73     while True:
  74         line = handler.next()
  75         if line.startswith('Predicted coding') or line is None: break
  76 
  77     fi = FastaIterator(handler)
  78     for gene in gr.genes:
  79         gene.setPeptide(fi.next())
  80         gene.setCds(fi.next())
  81     return gr